I am trying to use GCC to compile with nvcc as opposed to cl.exe using the command
nvcc -c -o depthconv_cuda_kernel.o depthconv_cuda_kernel.cu --compiler-bindir "C:\Program Files\mingw-w64\x86_64-8.1.0-posix-seh-rt_v6-rev0\mingw64\bin" --allow-unsupported-compiler -x cu -Xcompiler -fPIC -std c++11 --gpu-architecture=compute_86 --gpu-code=sm_86 -DGPU -I"C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.1/include" -DCUDNN --compiler-options "-DGPU -DCUDNN -fPIC"
Importantly I have set --compiler-bindir to the mingw bin directory, as well as used the flag --allow-unsupported-compiler (not sure if that is necessary).
I need gcc for the command line options std c++11 as well as -fPIC . It seems nvcc is still looking for MSVC's cl.exe, how can I change this so it uses the GCC compiler I need? I get the message:
Failed to run C:/Program Files/mingw-w64/x86_64-8.1.0-posix-seh-rt_v6-rev0/mingw64/bin/cl.exe (The system cannot find the file specified. ). nvcc fatal : Failed to preprocess host compiler properties.